Evaluation of WRF Modeling in Relation to Different Land Surface Schemes and Initial and Boundary Conditions: A Snow Event Simulation Over the Tibetan Plateau

Abstract

AbstractSnowfall and the subsequent evolution of the snowpack play important roles in the cryospheric and hydrospheric processes that occur on the Tibetan Plateau (TP). Current literature provides scarce evidence covering the sensitivity of solid precipitation to land surface physics schemes and initial and boundary conditions on the TP. Six numerical experiments using the Weather Research and Forecasting (WRF) model were conducted to simulate a snow event over the TP in March 2017. Different land surface physics schemes, that is, Community Land Model (CLM), Noah, and Noah‐MP, and initial and boundary conditions provided by atmospheric reanalysis data sets, that is, the National Centers for Environmental Prediction‐FNL and ERA‐Interim data sets, were applied in sensitivity analyses. The observed near‐surface air temperature, snow depth, and snow water equivalent (SWE) values were used to evaluate each model's performance. The results demonstrate that (1) the sensitivity of the near‐surface air temperature to land surface physics schemes is greater than it is to both the initial and boundary conditions; (2) the best performance is achieved when applying WRF + CLM with a root‐mean‐square error of 8.4 °C, a mean absolute deviation of 7.3 °C, a correlation coefficient of 0.75, and a spatial correlation coefficient of ~0.5 to air temperature estimates. A potentially important factor appears to be the advanced parametrization of albedo in the CLM scheme; (3) the advanced land surface schemes in the WRF model describes the physics of cryospheric and hydrospheric processes in detail, and the land surface response is determined by multiple variables and parameters in such schemes. The spatial patterns in such variables and parameters determined the detailed spatial variabilities observed in snow cover and amount and its temporal evolution. The WRF model overestimates, however, the intensity and extent of snow depth and SWE; (4) simulations of solid precipitation are more accurate when applying CLM or Noah‐MP + ERA‐Interim in WRF; and (5) WRF performance with regard to SWE estimates clearly depends upon the discrimination of lighter from heavier snowfall.